TorchToTMTensor CumSum Lowering Only Supports Constant Axis

Here,

https://github.com/llvm/torch-mlir/blob/34f6948533287e67801384cb00f7150edc1461a5/lib/Conversion/TorchToTMTensor/TorchToTMTensor.cpp#L1532

the TorchToTMTensor lowering checks to make sure that dim is a TorchConstantInt.

However, ONNX forces the axis to be variable because it allows axis to be a tensor.

This causes (at least part of) the lowering failure described in https://github.com/nod-ai/iree-amd-aie/issues/103

opt-125M.fp32.onnx.mlir:231:12: error: failed to legalize operation 'torch.aten.cumsum' that was explicitly marked illegal
    %228 = torch.operator "onnx.CumSum"(%226, %227) : (!torch.vtensor<[1,6],si64>, !torch.vtensor<[],si32>) -> !torch.vtensor<[1,6],si64>
           ^
opt-125M.fp32.onnx.mlir:231:12: note: see current operation: %287 = "torch.aten.cumsum"(%18, %286, %16) : (!torch.vtensor<[1,6],si64>, !torch.int, !torch.int) -> !torch.vtensor<[1,6],si64> loc("opt-125M.fp32.onnx.mlir":231:12)

Start at lib/Conversion/TorchToTMTensor/TorchToTMTensor.cpp around line 1532 and inspect the TorchToTMTensor lowering for torch.aten.cumsum. Compare its constant-axis check with the ONNX-generated operation and the opt-125M.fp32.onnx.mlir diagnostic. Done means the lowering accepts the variable axis form described here and no longer fails to legalize that cumsum operation.